Reproducibility, Source And Functional Analysis Of Mrna Signals In Cancer Peripheral Whole Blood

Many researchers try to develop biomarkers for cancer prevention, early detection and treatment by comparing cancer and normal peripheral whole blood gene expression profiles. However, few studies have examined whether the differential mRNA signals observed in cancer peripheral whole blood compared with normal samples for a cancer type are reproducible in different experiments. Moreover, as RNAs in peripheral whole blood mainly come from the blood leukocytes, it’s unclear whether such signals observed in cancer peripheral whole blood reflect gene expression changes of particular leukocytes under the cancer condition or the leukocyte subpopulation shifts. Even though the origin of differential mRNAs in cancer peripheral whole blood is uncertain, some researchers still perform the pathway enrichment analysis for biological interpretation of such blood-based differentially expressed genes between cancer and normal controls. However, the biological meanings underlying such genes are ambiguous. Therefore, in this dissertation, the reproducibility of differential mRNA signals identified from independent gene expression datasets for a same cancer is analyzed. Then, the influence of proportion changes of leukocyte subtypes on the cancer peripheral whole blood differentially expressed genes is investigated. Finally, the separate enrichment analyses of pathways for up- and down-regulated genes in cancer peripheral whole blood compared with normal controls are performed and their biological meanings are discussed.In chapter 2, using lung cancer as an example, we showed that the differentially expressed genes in lung cancer compared with normal samples respectively identified from three peripheral whole blood gene expression datasets were significantly reproducible. Then, by estimating the proportions of leukocyte subtypes in each sample, we showed that the proportions of myeloid lineage leukocytes tended to increase while the proportions of lymphoid lineage leukocytes tended to decrease under the lung cancer condition. By comparing the differentially expressed genes between myeloid and lymphoid lineage leukocytes to the differentially expressed genes between lung cancer and normal peripheral whole blood, we showed that the differential mRNA signals observed in lung cancer peripheral whole blood were mainly affected by the proportion changes of myeloid and lymphoid lineage leukocytes and reflected the expression differences between them. Similar analyses were done for three inflammation-associated pulmonary diseases including sarcoidosis, pneumonia and tuberculosis, which showed that the differentially mRNA signals observed in inflammation-associated pulmonary diseases were mainly influenced by the proportion changes of myeloid and lymphoid lineage leukocytes, reflecting the expression differences between them. The comparison of lung cancer differentially expressed genes to inflammation-associated pulmonary disease differentially expressed genes showed that the overlapping genes were 100% consistent in the sense of direction of regulation, hinting that the peripheral whole blood differential mRNAs in pulmonary cancer and inflammation-associated diseases were highly similar, both originated from the proportion changes of myeloid and lymphoid lineage leukocytes and reflected the differential signals between myeloid and lymphoid lineage leukocytes. Therefore, the comparison of cancer peripheral whole blood to normal blood may provide limited information for the development of cancer diagnostic biomarkers that could distinguish cancer from inflammation.In chapter 3, we summarized two strategies adopted in pathway enrichment analysis, and showed that the separate analysis of pathways for up- and down-regulated genes could identified more significant functional pathways than the analysis for all differentially expressed genes together based on five datasets of tumor gene expression profiles measured by microarray or RNA-seq. In chapter 4, we first showed that the differential mRNA signals observed in colorectal cancer compared with normal peripheral whole blood also originated from the leukocyte subpopulation shifts. Then, by performing functional enrichment analysis for up- and down-regulated genes in cancer peripheral whole blood samples compared to normal samples, we identified significant functional pathways that were commonly disturbed by lung cancer, colorectal cancer, breast cancer and ovarian cancer and showed that these functional pathways were all significantly enriched for the differentially expressed genes between myeloid and lymphoid lineage leukocytes, further suggesting that the differential mRNA signals observed in cancer peripheral whole blood mainly reflected the cell proportion changes in leukocytes rather than the dysfunction of some particular type of leukocytes.Therefore, the identification of peripheral whole blood based cancer biomarkers that could discriminate cancer from inflammation for clinical use should eliminate the influence of leukocyte subpopulation shifts or pay attention to some particular leukocyte subtypes.